Data-driven subgroups of adolescent depression: integrating EEG-derived brain networks, childhood trauma, and depressive symptoms

Background Adolescent major depressive disorder (MDD) is characterized by marked clinical heterogeneity and heightened sensitivity to psychosocial adversity, particularly childhood trauma. Ongoing neurodevelopment and brain plasticity during adolescence may limit the applicability of adult-derived neurobiological models. Previous studies have rarely integrated large-scale network connectivity, trauma exposure, and symptom dimensions within a single multivariate framework and have largely relied on functional magnetic resonance imaging or univariate electroencephalographic (EEG) measures. Methods This cross-sectional case-control study included 67 medication-free adolescents with MDD and 35 matched healthy controls. Resting-state source-space EEG functional connectivity was estimated. Sparse canonical correlation analysis (sCCA) was used to identify multivariate associations between network connectivity and clinical and childhood-trauma measures. Moderation analysis examined whether network connectivity moderated the association between trauma and depression severity. K-means clustering of sCCA-derived canonical variates was used to identify data-driven patient subgroups. Results Adolescents with MDD showed lower default mode network (DMN)-limbic network (LMN) connectivity than healthy controls. The only sCCA mode meeting the prespecified robustness criteria linked physical abuse, emotional neglect, and anxiety/somatization to distributed connectivity features, particularly those involving the frontoparietal, limbic, visual, and salience networks. Within-visual-network connectivity moderated the association between childhood trauma and depression severity. Clustering identified two patient subgroups with distinct symptom and network profiles. Conclusions EEG-derived network measures may help characterize the interplay between childhood trauma and clinical heterogeneity in adolescent MDD. These preliminary findings require replication in larger, independent samples. Clinical trial number not applicable